CAB and a Central Objec Model Design

Topics: CAB & Smart Client Software Factory
Feb 2, 2006 at 8:47 AM
originally posted by: abbrescia

I would like to know how CAB framework to fit/integrate with a central object model design like that described in the following article:

Thank you in advance.

Feb 3, 2006 at 8:16 AM
originally posted by: BradWilsonMSFT

I've put this article on the top of my "to read" stack and I'll let you know what I think. I will be interested to hear what others think as well. :)
Feb 7, 2006 at 6:18 AM
originally posted by: jhusain

I'm very interested in this topic as well. I was planning to use this approach before CAB came onto the scene. It seems to me that a central object model is antithetical to CAB which attempts to promote modularity (recommends giving each WorkItem it's own copy of state objects) while facilitating integration (using EventBroker).
Feb 11, 2006 at 8:06 PM
originally posted by: pwolf

Yes it would be very helpful to enable scripting in an application and still be able to have the benefit of using CAB.
So I am also very interested if someone has experience in bringing the two together.
Feb 20, 2006 at 10:35 AM
originally posted by: Lakushae

I am also interested. We started a new project following the Automation example in that MSDN article and are now trying to evaluate how to integrate CAB in it. Both designs are based on MVC both offer distinct complementary advantages.

Since the Automation Model is also a prerequisite to use Visual Studio Tools for Applications, it seems logical to see that it integrates nicely with CAB
Feb 20, 2006 at 3:00 PM
originally posted by: pwolf

I asked about it in a different thread:

and got this answer from Brad Wilson so far:
CAB neither helps nor hinders the addition of such an object model. It will be slightly challenging if you write a loosely coupled application. The solution chosen may depend on the requirements.

For example, you might choose to make a single object model available that covers everything in the application (shell & models), and design some kind of extensibility system for the object model so that new modules can plug into the system.

Or, you might choose to have a single object model per "logical" application (meaning, likely, one per module for any modules that might need to expose a model).

I agree though that I would find it very helpful if this becomes part of the CAB roadmap.